DETECTING ADVERSE WEATHER CONDITIONS AT A LiDAR SENSOR LEVEL

Abstract

Disclosed are systems and methods for detecting weather conditions at a LiDAR sensor level. In some aspects, a method includes calculating a reference probability mass function (PMF) of at least one field of a point cloud generated from reference scene responses received from a light detection and ranging (LiDAR) sensor; calculating a current PMF for the at least one field of the point cloud generated from a current scene response received from the LiDAR sensor; determining a statistical difference between the reference PMF and the current PMF using a Kullbeck-Leibler (KL) divergence calculation; and responsive to the statistical difference satisfying a threshold for the at least one field, flagging an environmental change in the current scene response.

What is claimed is: 
     
         1 . A method comprising:
 calculating a reference probability mass function (PMF) of at least one field of a point cloud generated from reference scene responses received from a light detection and ranging (LiDAR) sensor;   calculating a current PMF for the at least one field of the point cloud generated from a current scene response received from the LiDAR sensor;   determining a statistical difference between the reference PMF and the current PMF using a Kullbeck-Leibler (KL) divergence calculation; and   responsive to the statistical difference satisfying a threshold for the at least one field, flagging an environmental change in the current scene response.   
     
     
         2 . The method of  claim 1 , wherein the at least one field comprises a reflectivity field or a radial distance field. 
     
     
         3 . The method of  claim 1 , wherein the environmental change comprises an adverse weather condition. 
     
     
         4 . The method of  claim 1 , wherein the threshold is one of a plurality of thresholds, with each threshold of the plurality of thresholds corresponding to a particular weather condition. 
     
     
         5 . The method of  claim 4 , further comprising identifying the particular weather condition that corresponds to the threshold of the plurality of thresholds that is satisfied. 
     
     
         6 . The method of  claim 4 , wherein the particular weather conditions comprise at least one of light rain, moderate rain, heavy rain, sleet, hail, or snow. 
     
     
         7 . The method of  claim 1 , wherein the reference scene responses comprise an average of the at least one field over a plurality of previous scene responses. 
     
     
         8 . The method of  claim 1 , wherein the at least one field comprises a reflectivity field and a radial distance field, and wherein in response to the current PMF for the distance field being above the threshold corresponding to the distance field, skipping the flagging of the environmental change. 
     
     
         9 . The method of  claim 1 , wherein the at least one field comprises a reflectivity field and a radial distance field, and wherein in response to the current PMF for the distance field being below the threshold corresponding to the distance field, and the current PMF for the reflectivity field being above the threshold corresponding to the reflectivity field, flagging the environmental change. 
     
     
         10 . The method of  claim 1 , wherein the LiDAR sensor is comprised in an autonomous vehicle (AV). 
     
     
         11 . An apparatus comprising:
 one or more hardware processors to:
 calculate a reference probability mass function (PMF) of at least one field of a point cloud generated from reference scene responses received from a light detection and ranging (LiDAR) sensor; 
 calculate a current PMF for the at least one field of the point cloud generated from a current scene response received from the LiDAR sensor; 
 determine a statistical difference between the reference PMF and the current PMF using a Kullbeck-Leibler (KL) divergence calculation; and 
 responsive to the statistical difference satisfying a threshold for the at least one field, flag an environmental change in the current scene response. 
   
     
     
         12 . The apparatus of  claim 11 , wherein the environmental change comprises an adverse weather condition. 
     
     
         13 . The apparatus of  claim 11 , wherein the threshold is one of a plurality of thresholds, with each threshold of the plurality of thresholds corresponding to a particular weather condition, and wherein the one or more processors are further to identify the particular weather condition that corresponds to the threshold of the plurality of thresholds that is satisfied. 
     
     
         14 . The apparatus of  claim 11 , wherein the at least one field comprises a reflectivity field and a radial distance field, and wherein in response to the current PMF for the distance field being above the threshold corresponding to the distance field, the one or more processors are to skip the flagging of the environmental change. 
     
     
         15 . The apparatus of  claim 11 , wherein the at least one field comprises a reflectivity field and a radial distance field, and wherein in response to the current PMF for the distance field being below the threshold corresponding to the distance field, and the current PMF for the reflectivity field being above the threshold corresponding to the reflectivity field, the one or more processors are to flag the environmental change. 
     
     
         16 . A non-transitory computer-readable medium having stored thereon instructions that, when executed by one or more processors, cause the one or more processors to:
 calculate a reference probability mass function (PMF) of at least one field of a point cloud generated from reference scene responses received from a light detection and ranging (LiDAR) sensor;   calculate a current PMF for the at least one field of the point cloud generated from a current scene response received from the LiDAR sensor;   determine a statistical difference between the reference PMF and the current PMF using a Kullbeck-Leibler (KL) divergence calculation; and   responsive to the statistical difference satisfying a threshold for the at least one field, flag an environmental change in the current scene response.   
     
     
         17 . The non-transitory computer-readable medium of  claim 16 , wherein the environmental change comprises an adverse weather condition, and wherein the threshold is one of a plurality of thresholds, with each threshold of the plurality of thresholds corresponding to a particular weather condition. 
     
     
         18 . The non-transitory computer-readable medium of  claim 17 , wherein the one or more processors are further to identify the particular weather condition that corresponds to the threshold of the plurality of thresholds that is satisfied. 
     
     
         19 . The non-transitory computer-readable medium of  claim 16 , wherein the at least one field comprises a reflectivity field and a radial distance field, and wherein in response to the current PMF for the distance field being above the threshold corresponding to the distance field, the one or more processors are to skip the flagging of the environmental change. 
     
     
         20 . The non-transitory computer-readable medium of  claim 16 , wherein the at least one field comprises a reflectivity field and a radial distance field, and wherein in response to the current PMF for the distance field being below the threshold corresponding to the distance field, and the current PMF for the reflectivity field being above the threshold corresponding to the reflectivity field, the one or more processors are to flag the environmental change.